The overall goal of this application is to elucidate genetic risk modifiers influencing age of onset of hereditary nonpolyposis colorectal cancer (HNPCC) using a pathway-based approach. The overall hypothesis of this proposal is that SNPs in cell cycle pathways influence the efficiency of the finely tuned mechanisms of cell cycle progression. We will continue to build and expand upon our unique clinical and specimen resource of mismatch repair gene mutation carriers in order to accomplish this goal. In aim 1, we will determine the role that genetic variants in cell cycle genes have on risk for development of early onset HNPCC. We will use the Illumina Golden Gate platform to genotype SNPs from 153 cell cycle genes. We will evaluate the main effects of SNPs in these genes using a two-phase study design in order to determine if there are differences in age-associated risk for development of HNPCC between carriers of the wild type genotype vs. those carrying one or two copies of the SNP. In aim 2, we will use novel machine learning analytic tools to determine if we can identify combinations of cell cycle SNPs that work together to influence risk for development of early onset HNPCC. Two approaches will be used to accomplish this goal. A focused candidate gene approach, and the second is an exploratory, pathways based approach. In the focused candidate approach, we have selected 12 candidate genes because they are known to interact with each other in the cell cycle pathways and there is evidence that SNPs in these genes influence risk for HNPCC or other cancers. We will determine if combinations of these SNPs work together to influence development of HNPCC at an early age. In the second approach we will investigate SNPs in all 153 of the genes, including the 12 candidate genes. We hypothesize that there are many possible combinations of cell cycle genes that work together and under the influence of SNPS, contribute to development of HNPCC at an early age. We will use Random Forests (RF) modeling to rank all of the SNPs from these genes. RF can handle large numbers of predictor variables such as SNPs and provides estimates of variable importance. We will then use CART to build age-to-onset trees using the SNPs ranked the highest from RF. The proposed studies will contribute to our long-term goal to develop a risk model for early onset HNPCC by defining the role that SNPs in cell cycle genes play in risk for HNPCC. They will also provide important information regarding novel gene-gene interactions contributing to risk for development of HNPCC at an earlier age. The National Cancer Institute in its 2006 budget proposal cited risk prediction as an area of extraordinary opportunity. Risk prediction models for cancer could provide valuable tools for identifying individuals who may benefit from preventive treatments or increased surveillance or who are good candidates to participate in clinical trials. PUBLIC HEALTH REVELANCE: The proposed studies will contribute to our long-term goal to develop a risk model for HNPCC, which may also have important implications for sporadic colorectal cancer as well as other cancer sites. [unreadable] [unreadable] [unreadable] [unreadable]